Re: [PATCH V2 4/4] mfd: mediatek: add MT6323 support to MT6397 driver

[Lee Jones]

On Tue, 26 Jan 2016, John Crispin wrote:

> 
> 
> On 26/01/2016 09:34, Lee Jones wrote:
> > On Tue, 26 Jan 2016, John Crispin wrote:
> >> On 26/01/2016 04:07, Henry Chen wrote:
> >>> On Mon, 2016-01-25 at 19:59 +0100, John Crispin wrote:
> >>>>
> >>>> On 25/01/2016 19:44, Matthias Brugger wrote:
> >>>>> On Monday 25 Jan 2016 16:36:40 John Crispin wrote:
> >>>>>> Hi,
> >>>>>>
> >>>>>> On 25/01/2016 13:41, Lee Jones wrote:
> >>>>>>> Please honour the subject format of the subsystem you are contributing
> >>>>>>> to.
> >>>>>>>
> >>>>>>> `git log --oneline -- $subsystem` gives you this.
> >>>>>>>
> >>>>>>> On Mon, 25 Jan 2016, John Crispin wrote:
> >>>>>>>> Signed-off-by: John Crispin <blogic@xxxxxxxxxxx>
> >>>>>>>> ---
> >>>>>>
> >>>>>> [...]
> >>>>>>
> >>>>>>>> @@ -261,6 +271,15 @@ static int mt6397_probe(struct platform_device
> >>>>>>>> *pdev)
> >>>>>>>>
> >>>>>>>>  	}
> >>>>>>>>  	
> >>>>>>>>  	switch (id & 0xff) {
> >>>>>>>>
> >>>>>>>> +	case MT6323_CID_CODE:
> >>>>>>>> +		mt6397->int_con[0] = MT6323_INT_CON0;
> >>>>>>>
> >>>>>>> This is confusing.  You're still using memory allocated for a mt6397
> >>>>>>> device.
> >>>>>>
> >>>>>> the variable is currently defined as struct mt6397_chip *mt6397;
> >>>>>> shall i only change the name or also create a patch to rename the struct ?
> >>>>>>
> >>>>>
> >>>>> I think we should rename the struct and the file as well.
> >>>>>
> >>>>> Cheers,
> >>>>> Matthias
> >>>>
> >>>> Hi,
> >>>>
> >>>> that would have been my next question. renaming the struct would imply
> >>>> renaming the driver and the whole namespace contained within. We would
> >>>> then also need to change the Kconfig and Makefile. I am happy to do this
> >>>> but want to be sure that is is actually wanted.
> >>>>
> >>>> 	John
> >>> Hi,
> >>>
> >>> Since mt6323 was similar with mt6397, I think we can reuse the
> >>> mt6397_chip without duplicate code.
> >>>
> >>> Maybe we can rename the local variable name to avoid confusing.
> >>>
> >>> struct mt6397_chip *mt_pmic;
> >>> ...
> >>> ...
> >>> switch (id & 0xff) {
> >>> 	case MT6323_CID_CODE:
> >>> 		mt_pmic->int_con[0] = MT6323_INT_CON0;
> >>> 		mt_pmic->int_con[1] = MT6323_INT_CON1;
> >>> ...
> >>> ...
> >>>
> >>> Henry
> >>
> >> Hi,
> >>
> >> IMHO we should either rename the namespace or not. renaming some
> >> variables seems weird as that will just move the confusion/inconsistency
> >> to another place in the code. I am however rather indifferent on this
> >> matter.
> > 
> > It's common to name a driver after the device which was enabled first,
> > so no need to rename the files or CONFIGs; however, it does seem
> > prudent to generify the struct (both parts).
> > 
> 
> Hi,
> 
> renaming struct mt6397_chip is not as straightforward as it seems as
> other drivers such as pinctrl and rtc share the data structure with the
> mfd driver.
> 
> i have kept the structs name as is for now and only renamed the instance.

That's fine.

> If you want i can create a separate series that addresses only the
> renaming of the structure. The "rename struct mt6397" patch would need
> to do a change to drivers from 3 subsystems in 1 single commit as not to
> break compile and merge order.

I'll leave this up to you.

The real problem from my PoV was the instance name.

-- 
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog